What can we do?

Population Characteristics and the Sigma Tool For Change
Harry Rosenberg
Aug, Oct 2004

Discussion of definitive features with an eye on the war on terrorism

Extremism infects and affects many walks of life. And it expresses itself in various ways. Extremism is what the word implies, behaviors at the extremes farthest removed from the middle ground. Attributes of critical importance on this web site are governance (usually expressed as politics), freedom (or liberty, a similar concept), economics (individual and national wealth), and the like. Consider the following table:

Extremism in Practice

Attribute	A Extreme	B Extreme
Governance	Total Top-Down Control	Total Bottom-Up Control; Anarchy
Freedom	No Individual Freedom	Total Individual Freedom
Economics	Penniless	Individual Wealth Without Bound
Ethnic Attitude	Total Racism and Bigotry	Total Acceptance of All Others
Religion	Fundamentalism	Atheism
Violence	War, Terror, Genocide	Peace

From this simple example it is easy to see why societies need controls on those things affecting one's position in life as well as how well a person fits into that position. The most extreme positions do not usually support viable societies. There is irony on that very point, however. Monotheism is some 35 times more violent than is atheism at the present time. Eastern religions are 22 multiples more peaceful than monotheism. See Religion and Violence for more on that. Tables like the above define extremism; what is good for society lies in the middle ground, between the extremes.

We need a new word(s) to describe the above M x m table that defines extremism. UMBRA- (Latin for shade) is a possibility. We might use the politically correct word, moderate, except that it is subjective, where objectivity is needed. For most of a millennium now, struggles have been mounted against top-down governance to the point where freedom and liberty are now enjoyed by much of the world's population as a desirable condition. The march of Science and its social companion, technology, has vastly improved the well being of most of the earth's people. Yet huge disparities remain. Fundamentalism has been around since Abraham. Moses produced the ten commandments in an attempt to bring moral order to the seeming moral anarchy of the times.

Like the crusading popes and monarchs of the middle ages, extremism is still expressed politically in much of the world. And like most societies, most people on earth fall somewhere in the middle ground on each of the above scales.


Those who would trade liberty for security deserve neither.-Ben Franklin-

Franklin raises a vital point. In keeping with our philosophy on this site, we believe that even here, there is a middle ground. We see no practical way to provide a level of liberty that amounts to total self determination for every individual on earth--it would be anarchy given our genetic makeup. Nor do we see total control as a feasible alternative.

There can be any number of rows. But each row must have an attribute that has has two genuine extremes that one can not go beyond. This can quickly get quite complex. 

Scale is another issue.  A scale might be a degree of totality on a 0 - 100 or 0 - 30 scale. Economics in a relative sense could be numeric. But how does one give numerical measure to something as vague numerically as religion or governance?

In pure mathematics there are such things as boundary conditions. For example, the sine and cosine functions in trigonometry can only have values between 1 and -1. And zero as a divisor is a boundary of another sort; all numbers larger or smaller are permitted, but zero is not in any practical sense.

ZEMPLAR attributes that are not measurable in a numeric sense, can nevertheless be handled by statistical mathematics in a semi quantitative way. This procedure in and of itself will not define an individual person with high accuracy. But it will define groups of people with sufficient accuracy to predict their behaviors to useful degrees of precision with respect to an expectation. There are two approaches. One is to apply non-parametric procedures and the other is to devise scale scores that yield numerical ratings. The latter has a long and valid history of application and is well understood. Since people and their behavior are involved, such scales have become known as psychometrics. A psychometric test is usually made up of several scale scores. And a battery of suck tests is more definitive than any test alone would be. In our nation's history, such testing has served us well--when used properly of course. For example, during the Cold War, workers assembling nuclear bombs were given a battery of psychometric and other tests to assess their likelihood of doing good jobs handling critical material. No failures of the procedure are known.

In statistical math, the standard deviation, SD, which is easily calculated, is the key.  SD is also known as sigma. Sigma provides the basis for the quality control systems that developed in Japan and which gave Detroit a long term migraine. Sigma has proven its veracity in many fields, including astrophysics, medicine, biology, and psychometry. It defines the inflection point of a bell curve and underlies all parametric statistical tests.

Here is how Sigma works. About 2/3 of a population of anything is within about one standard deviation of the overall average for any given attribute. About 1/20 of the population is within two standard deviations, and one in 370 are within about three standard deviations of the grand average.  Sigma provides a way of thinking, and provides possible solutions to many disparate problems. The key concept is that sigma describes something called the distribution. When the math requirements are all met, sigma can provide astonishing accuracy in the description of a distribution and how a population behaves in the large. That is where the 2/3, 1/20 and 1/370 fractions come in. Applied to ethnic attitude for example, we would agree that anyone in the 2/3 group might be considered typical, because they are indeed typical. The 1/370 group might well deserve the label of being extreme, simply because they are so rare and far removed from the rest of us.

The problem is that sigma requires numerical data to calculate.  Individual net worth would provide sigma from economic measures, say individual net worth. But how can we apply sigma to ethnic attitude?

People who make psychometric testing their business, have solved this problem. They are able to develop scale scores that estimate quite well where a person sits or lies overall on the scales that are not entirely objective.  A professionally given IQ test is one such scale score composed of performance in several dimensions of intelligence, some more objective than others.

Psychology is necessarily a softer "science" than physics.  But that does not mean a shadow is not better than nothing, it surely is; it is just less definite.

For example, here is how it can work for a governance scale involving gender and business issues.

Select the one item that best describes your belief:

I believe in:

• Total freedom--Anarchy.
• Total freedom to do as I please except murder anyone.
• Total freedom to do as I please except the 10 commandments.
• Freedom to do as I please except for those curbs necessary to maintain local law and order.
• The same as 4 except that women must defer to men.
• The same as 7 except that women be allowed to govern.
• Total dictatorship by males except that business should not have to pay corporate income taxes.
• The same as 7 excepts that I should not have to pay corporate income taxes.
• The same as 8 except that I be allowed to govern my company as I see fit with freedom from antitrust law.
• Total dictatorship by males in all walks of life.

A psychometric test has several scales of this type that reflect attitudes or traits that produce behavior of one sort or another. In the above scale, much, maybe most, of people in the world societies today would score roughly between 4 and 6. Scoring in the 1-3 or 7-10 range might be considered extremism in this model. A more valid set of questions, with a wider blue area, would seem desirable. And we agree--except for one thing. 65-80% of all of us in America at least, have Authoritarian Personalities!

When taking a test such as this, a person can only check off one statement, and that statement has a number attached to it.  If 100 people took this test an average and standard deviation with good validity can be calculated from the scores.  From there you are off to the races, not with precision, just with a probability of 5, 10, or even a 20 times more likely to be right those 100 people than would the typical politician. And they will hate you for it!

All such psychometric tests assume the respondent is truthful, and even more important, that the respondent is self aware enough to know the answer.  Nevertheless the technique is powerful. 

For further example, an ethnic-attitude scale could range between say total acceptance of others or total bigotry (non-acceptance). The psychometrist would then devise a series of say at least 30 questions; this also defines the number of columns in our matrix. The questions could grade from single questions whose answers would be indicative of absolute acceptance or absolute non-acceptance, leaving 28 questions available to dealing with and defining the various degrees of acceptance nearly all of us have. For a simplified example of how numbers are derived from questions asked, if you answered half of the questions on ethnic tolerance yes and the other half no, you would have a scale score of 50 on a scale of one to a hundred. Your scale scores over ten attributes might range from 39 to 66. Altogether, the individual scale scores would provide a profile of your tendency toward extremism. These numbers would not be extreme for a standard deviation of 16 on each scale for example. We might begin to worry about people below 10 or above 90, and worry a lot about those showing two or more extreme scales or anyone approaching say 3 or 97 on any such scale.

Although the foregoing looks formidable to those averse to math, such procedures lie at the heart of the scientific method.

The above examples are illustrative only; we are not ready to recommend policy here. Even so, using this general approach in defining traits associated with violence, and using the results to ease conditions leading to violence, beats anything we are doing now. Basically, it is no more than doing the things that work, fopr example in Honolulu, El Paso, Lisbon, Oslo, and Tokyo. At the same time we can avoid the features that make Washington DC, Baltimore and Detroit such violent cities.

In operation, if your UMBRA were 56 +/- 8, and mine was 15 +/- 5, I would be regarded as being more extreme than you. We would belong to different populations of traits along the bell curve. If a society is serious, it could look at itself in this or similar ways and work to remove the social reasons for extremely high or low scores. Is our societry serious? Not yet we fear.

Think about it. We kill more people with guns than the terrorists have to date. This is a fact. By "attacking" (a terrible word to use) violence, instead of just its terror component, we can give new meaning to the word PEACE.

One good feature of psychometrics is that such tests can have ten or more scales altogether and be completed in a couple of hours.  The worst part of it is that to be valid, those applying or taking them either must have faith in the system and be open and strong enough not to fear misuse. This is one barrier.  Another is authoritarian governors who fear putting such knowledge to use in achieving peace; they prefer conflict and need someone to look down upon, never mind that Humiliation leads to violence; they will find numerous reasons for not using the sigma technique.

Who would not be willing to lose this bit of privacy to make progress in the war on terror?  Psychometry merely labels behavior; It does not intrude on matters of real privacy, like why are we the way we are? It merely describes personality or behavior expressed that others see anyway.

Some people fear psychometry with justification. For example, IQ test results have been applied improperly to whole populations. William Shockley did exactly that in attempting to show Caucasians superior to Blacks. His approach showed Asians to be brighter still, a fact he chose not to highlight. His assumption (apparently driven by bigotry) that the IQ test is free of cultural bias was his basic error; it is not. An IQ test is not a psychometric test. IQ predicts how well a person will do in school. Shockly went beyond the bounds of good science reporting. His claims supported bigotry and set back integration. Psychometric tests are designed to reveal personality features and attitudes such as bigotry, friendliness, sociability and so on. The potential for misuse is there and must always be guarded against.

So traditional and new safeguards for individual privacy for all those involved must be enforced rigorously. Improper use of psychometric information by any person other than the testee must have certain and serious consequences. New safeguards could include double blind testing where the testees are anonymous to the tester and only the statistics for the test cohort are ever released to the public. Once the statistics are certified, the actual test would be destroyed. There are numerous other measures to protect individual privacy.

There will be many individuals not attuned to these procedures for various reasons. They need not take part as there are many who will. If a large fraction of a selected population refused to participate, a study would take longer with added costs to establish validity. This is because the sample would no longer be truly random and would need to be larger. Each such refusals might require two individuals replace one refusal to maintain scientific validity.

On this point, a reviewer of this page made a statement, then asked a question:

"Privacy is a fundamental liberty recognized by every court in the land to at least some degree. Would not giving up a little privacy be giving up some fundamental liberty?"

The answer is of course, at least in a sense. Liberty is defined as the rights one has in a society. Freedom is often used synonymously. But the rights we have can never be absolute--we do not after all behave like animals. So we have already lost some liberty or freedom. Psychometry for the sake of peace seems to fall in the same class as the liberties we have already given up--mostly without even realizing it.

A final note to the theoretically inclined. Each scale of a profile in a psychometric test has its own standard deviation for a population.  To be useful in assessing extremism, the distribution must ideally fit within the boundary values, the extremes. For example, how do we handle a scale where sigma is so large that it puts some people off scale theoretically? Such an event would simply prove the so-called bell curve does not describe data of this type.

That need not be fatal as there are two approaches to this situation. The first one is simply to ignore the fact, and that is what too many of the users of sigma do--mostly in innocence, but that should no happen here. And more often than not in such cases, little or no damage is done. But there can be serious exceptions. The other approach is to find a distribution function that fits the data and remains within the scale limits and use it. That feature may require that the data be transformed by a formula that stays within bounds. The details are too arcane to describe here, but finding a suitable transform is duck soup for most statisticians. So also for interpreting the data. The proposal above is not the only possible design; it may not even be best. But it is a start.

There you have some of the details. The society that ignores the many opportunities to use the sigma tool to learn about the origins of violence and its control is risking its very future. So how do we go about using this tool?

Let's take Washington DC and Honolulu for one example; the former is some 23 times more violent than the latter. There is a reason. With sigma, we can devise any number of psychometric tests from which we can discover the meaningful correlation or correlations between the variables studied and the degree of violence. In fact, two such causes have already been identified to a rather high degree of likelihood, Ethnic Integration and Monotheism. The former was a formal experiment design, the latter lacked scientific rigor, but is so overwhelming, and in such agreement with numerous other observers, that it too has a high likelihood. Ethnic integration will go far toward addressing each of these causative factors.

Once in place, some years may be needed to validate a social effect on violence in a new setting, especially if the populations compared are not very different or changes enacted to make comparisons take place slowly.

As is well-known, samples selected randomly to be representative of the population can provide a picture of that population. One appropriate scale on a psychometric test could be violence itself. Another, from Varshney's work would be ethnic acceptance. Still others could include religiosity and Authoritarianism. Honolulu as a city reflecting its citizens would surely have a profile different from Washington DC. With that information in hand one could devise strategies to "peacify" Washington DC.

Only a politician would presume he already knows all these answers and toss the sigma experiment into the nearest trash can. The Bush Administration has already done exactly that. One need look no further than the sub-par performance of Charter Schools created under "No Child Left Behind." His words of description sound great, but Bush knows nothing about education. At the least, he should have had others apply the sigma tool before leaping into unknown waters on such a grand style.

The bright side is that studies of this type need only be done enough times to validate an effect. And it need not be done on many people either. But it might well take a decade or more to flesh out all the significant variables.

Evolving a social order that effectively removes violence from the radar screen will certainly take time also. But as soon as one variable is identified, and two already have been--work toward a new order can begin. It may well take half a century or more to bring peace to all humanity, given that poverty and population growth scale together.

Most basically, the Sigma Tool can tell you what is not real and what is likely to be real. In each case it can assign a probability to there being no effect; The sigma tool cannot do the opposite, however.

The technology for moving toward peace is no problem; motivating nations, smaller political units, and people to proceed seriously is a larger issue.

An objection surely to be raised is cost. Yes there is a significant cost. But in perspective, it is minor. Think of the national census, theoretically reaching every citizen. Compare that cost with two or three similar efforts in only a dozen cities. Compared with Iraq, the cost is peanuts. Compared with Iraq, it would be gigantic in effectiveness.

What are we waiting for?